LSTM-based dialysis fistula maturity detection method, system, device and medium

Through the method based on LSTM neural network, a sound classification model was established, which solved the accuracy of fistula maturity detection in hemodialysis, achieved automated and accurate detection results, and reduced the pain to patients and the cumbersomeness of the detection process.

CN116763265BActive Publication Date: 2024-11-15PEOPLES HOSPITAL OF HENAN PROV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310723197.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-06-16
Publication Date
2024-11-15
Estimated Expiration
2043-06-16

AI Technical Summary

Technical Problem

In the prior art, the maturity detection of fistula in patients with hemodialysis relies on manual examination, which is cumbersome and inaccurate, causing pain in the patient, and the speech analysis relies on subjective standards, and lacks effective detection methods.

Method used

Using an LSTM neural network method, a sound classification model is established by collecting sound samples, preprocessing and feature extraction, and a sound classification model is established to distinguish between normal and abnormal fistula sounds and improve detection accuracy.

Benefits of technology

Through the automated LSTM model, the accuracy of fistula maturity detection is improved, manual intervention is reduced, the cumbersomeness and pain of the detection process are reduced, and more reliable diagnostic results are provided.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116763265B_ABST
    Figure CN116763265B_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of sound classification, and discloses a method, system, device and medium for detecting the maturity of dialysis fistula based on LSTM, wherein the method comprises a collection module collecting sound samples and establishing a sound sample database; a feature extraction module preprocessing the sound samples and performing feature extraction to obtain feature parameters; a training module training based on a preset LSTM neural network and the feature parameters to establish a sound classification model; a test module tests the target sample through the sound classification model and provides a classification result. In the above manner, after the present application establishes a sound sample database through the collection module, the feature extraction module extracts the features of the training set and trains the features of the training set through the training module, and after the sound classification model is established, the test module classifies the target sample, thereby improving the detection accuracy of the maturity of fistula in hemodialysis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of sound classification, and in particular to a dialysis fistula maturity detection method, system, device and medium based on LSTM. Background Art

[0002] In the medical field, for hemodialysis patients, in order to prepare for hemodialysis, a superficial blood vessel, namely an arteriovenous fistula, needs to be established. Pathological sound is the sound caused by abnormalities in the vocal system, and the sound characteristics of arteriovenous fistula are pathological sound characteristics.

[0003] At present, the detection and evaluation of pathological sounds of such patients still mainly rely on manual inspection, which is a cumbersome process and often causes extreme pain to patients. In addition, the clinical analysis of speech mostly adopts subjective standards, which has many unfavorable factors. Therefore, how to improve the detection accuracy of fistula maturity in hemodialysis has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] The present application provides a LSTM-based dialysis fistula maturity detection method, system, device and medium to improve the detection accuracy of fistula maturity in hemodialysis.

[0005] In a first aspect, the present application provides a LSTM-based dialysis fistula maturity detection method, the method comprising:

[0006] The acquisition module collects sound samples and establishes a sound sample database;

[0007] The feature extraction module preprocesses the sound sample and extracts features to obtain feature parameters;

[0008] The training module performs training based on the preset LSTM neural network and the characteristic parameters to establish a sound classification model;

[0009] The testing module tests the target sample through the sound classification model and provides a classification result.

[0010] Furthermore, the feature extraction module pre-processes the sound sample, including:

[0011] The sound sample is pre-emphasized by a high-pass digital filter, and the preset transfer function of the high-pass digital filter is H(z)=1-az -1 , the relationship between the output of the pre-emphasis network and the input speech signal s(n) is:

[0012]

[0013] Where a is the pre-emphasis coefficient.

[0014] Furthermore, the feature extraction module pre-processes the sound sample, including:

[0015] Based on the short-term stability of the sound sample, the feature extraction module performs a short-term analysis of the sound sample by using a weighted method using a movable window of finite length, wherein the number of frames of the sound sample is f n =(N-overlap) / inc=(N-wlen) / inc+1, wlen is the frame length, inc is the frame shift, N is the length of the voice data, and overlap is the overlapping part.

[0016] Furthermore, the feature extraction module pre-processes the sound sample, including:

[0017] Based on a preset window function, the feature extraction module performs windowing processing on the sound sample, wherein the preset window function is:

[0018]

[0019] Furthermore, feature extraction is performed to obtain feature parameters, including:

[0020] The characteristic parameters are extracted by fast Fourier transform DFT, wherein the DFT is

[0021]

[0022] (n) is the input speech signal, and N represents the number of Fourier transform points;

[0023] The characteristic parameters of the sound sample are extracted by Mel filter bank, logarithmic operation, discrete cosine transform and / or dynamic feature extraction method.

[0024] Furthermore, the training module is trained based on the preset LSTM neural network and the characteristic parameters, before establishing the sound classification model, including:

[0025] The training module builds a preset number of LSTM units, connects a Dropout layer after each LSTM unit, predicts the results through a Softmax layer and obtains a loss function;

[0026] After obtaining the loss function, the loss function is optimized by the Adam optimization algorithm to find the global optimal point and introduce quadratic gradient correction.

[0027] Furthermore, the testing module tests the target sample through the sound classification model and provides classification results, including:

[0028] The target sample is input into the sound classification model for testing, and the result is output as 0 and 1, where 0 represents normal fistula sound and 1 represents abnormal fistula sound.

[0029] In a second aspect, the present application also provides a LSTM-based dialysis fistula maturity detection system, the system comprising:

[0030] A collection module, used for collecting sound samples and establishing a sound sample database;

[0031] A feature extraction module is used for preprocessing the sound sample and extracting features to obtain feature parameters;

[0032] A training module, used for training the training module based on a preset LSTM neural network and the characteristic parameters to establish a sound classification model;

[0033] The testing module is used to test the target sample through the sound classification model and provide classification results.

[0034] In a third aspect, the present application also provides a computer device, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the above-mentioned LSTM-based dialysis fistula maturity detection method when executing the computer program.

[0035] In a fourth aspect, the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement the LSTM-based dialysis fistula maturity detection method as described above.

[0036] The present application discloses a LSTM-based dialysis fistula maturity detection method, system, device and medium, wherein the LSTM-based dialysis fistula maturity detection method includes an acquisition module that acquires sound samples and establishes a sound sample database; a feature extraction module that preprocesses the sound samples and extracts features to obtain feature parameters; a training module that trains based on a preset LSTM neural network and the feature parameters to establish a sound classification model; and a test module that tests the target sample through the sound classification model to provide a classification result. In the above manner, after the present application establishes a sound sample database through the acquisition module, the feature extraction module extracts the features of the training set and trains the features of the training set through the training module, and after the sound classification model is established, the test module classifies the target sample, thereby improving the detection accuracy of the maturity of the fistula in hemodialysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 is a schematic flow chart of a LSTM-based dialysis fistula maturity detection method provided in the first embodiment of the present application;

[0039] Figure 2 is a schematic flow chart of a LSTM-based dialysis fistula maturity detection method provided in the second embodiment of the present application;

[0040] Figure 3 A schematic block diagram of a LSTM-based dialysis fistula maturity detection system provided in an embodiment of the present application;

[0041] Figure 4 A schematic block diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0043] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0044] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0045] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0046] The embodiments of the present application provide a method, system, device and medium for detecting the maturity of dialysis fistula based on LSTM. The method for detecting the maturity of dialysis fistula based on LSTM can be applied to a server. After the sound sample database is established by the acquisition module, the features of the training set are extracted by the feature extraction module and the features of the training set are trained by the training module. After the sound classification model is established, the target sample is classified by the test module, thereby improving the detection accuracy of the maturity of fistula in hemodialysis. The server can be an independent server or a server cluster.

[0047] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0048] See also Figure 1 , Figure 1 This is a schematic flow chart of a LSTM-based dialysis fistula maturity detection method provided by the first embodiment of the present application. The LSTM-based dialysis fistula maturity detection method can be applied to a server, and is used to establish a sound sample database through an acquisition module, extract the features of a training set through a feature extraction module, and train the features of the training set through a training module, and after establishing a sound classification model, the test module classifies the target sample, thereby improving the detection accuracy of fistula maturity in hemodialysis.

[0049] like Figure 1 As shown, the LSTM-based dialysis fistula maturity detection method specifically includes steps S10 to S40.

[0050] Step S10: The acquisition module acquires sound samples and establishes a sound sample database;

[0051] Step S20: The feature extraction module preprocesses the sound sample and extracts features to obtain feature parameters;

[0052] Step S30: The training module performs training based on the preset LSTM neural network and the characteristic parameters to establish a sound classification model;

[0053] Step S40: The testing module tests the target sample through the sound classification model and provides a classification result.

[0054] In a specific embodiment, before the feature extraction module analyzes and processes the labeled sound sample database, it must perform pre-processing operations such as pre-emphasis, framing, and windowing to eliminate the impact of aliasing, high-order harmonic distortion, high frequency, and other factors caused by the device collecting the voice signal on the quality of the voice signal. It is possible to ensure that the signal obtained by subsequent voice processing is more uniform and smooth, provide high-quality parameters for signal parameter extraction, and improve the quality of voice processing.

[0055] When the acquisition module establishes the sound sample database, it first converts the format of the sound file into the wave file format and then marks the sound samples, marking normal sound samples as 0 and abnormal sound samples as 1.

[0056] The present embodiment discloses a LSTM-based dialysis fistula maturity detection method, system, device and medium, wherein the LSTM-based dialysis fistula maturity detection method includes an acquisition module that acquires sound samples and establishes a sound sample database; a feature extraction module that preprocesses the sound samples and performs feature extraction to obtain feature parameters; a training module that trains based on a preset LSTM neural network and the feature parameters to establish a sound classification model; and a test module that tests the target sample through the sound classification model to provide a classification result. In the above manner, after the present application establishes a sound sample database through the acquisition module, the feature extraction module extracts the features of the training set and trains the features of the training set through the training module, and after the sound classification model is established, the test module classifies the target sample, thereby improving the detection accuracy of the maturity of the fistula in hemodialysis.

[0057] based on Figure 1 In the embodiment shown, in this embodiment, before step S20, the following steps are included:

[0058] The sound sample is pre-emphasized by a high-pass digital filter, and the preset transfer function of the high-pass digital filter is H(z)=1-az -1 , the relationship between the output of the pre-emphasis network and the input speech signal s(n) is:

[0059]

[0060] Wherein, a is the pre-emphasis coefficient, which is generally 0.9-1.0. In this feature extraction module, a=0.98 is taken.

[0061] Based on the short-term stability of the sound sample, the feature extraction module performs a short-term analysis of the sound sample by using a weighted method using a movable window of finite length, wherein the number of frames of the sound sample is f n=(N-overlap) / inc=(N-wlen) / inc+1, wlen is the frame length, inc is the frame shift, N is the length of the voice data, and overlap is the overlapping part.

[0062] Specifically, since the sound signal has short-term stability, the speech signal can be considered to be approximately unchanged within 10-30ms, and the sound signal can be divided into several short segments for processing. Therefore, after the pre-emphasis of the internal fistula sound signal is completed, it is framed.

[0063] The feature extraction module performs short-term analysis on the pre-emphasized sound signal. After the pre-emphasis of the fistula pathology sound signal is completed, the feature extraction module performs framing. The step S2 implements the framing of the speech by using a movable finite-length window for weighting. The frame length set in the present invention is 25ms and the frame shift is 10ms.

[0064] based on Figure 1 In the embodiment shown, in this embodiment, before step S20, the following steps are included:

[0065] Based on a preset window function, the feature extraction module performs windowing processing on the sound sample, wherein the preset window function is:

[0066]

[0067] based on Figure 1 In the embodiment shown, in this embodiment, step S20 includes:

[0068] The characteristic parameters are extracted by fast Fourier transform DFT, wherein the DFT is

[0069]

[0070] (n) is the input speech signal, and N represents the number of Fourier transform points;

[0071] The characteristic parameters of the sound sample are extracted by Mel filter bank, logarithmic operation, discrete cosine transform and / or dynamic feature extraction method.

[0072] Specifically, the feature extraction module passes the fast Fourier transformed fistula sound signal through a set of Mel-scale triangular bandpass filters to smooth the spectrum, eliminate the effect of harmonics, and highlight the resonance peak of the original sound sample. The relationship between the Mel scale and the frequency f is:

[0073]

[0074]

[0075] The frequency response of a triangular filter is defined as:

[0076]

[0077] After the energy spectrum passes through the Mel filter, the logarithmic energy of each filter group output is calculated to convert the convolution signal into an additive signal. The formula is as follows:

[0078]

[0079] Furthermore, the feature extraction module brings the above logarithmic energy into discrete cosine transform, converts it into the pitch information and channel information separation of the additive signal, and calculates the L-order Mel-scale Cepstrum parameter, the formula is as follows:

[0080]

[0081] Wherein M is the number of triangular filter banks, and the present invention takes the value of 24. L order refers to the order of MFCC coefficients, which is usually 12-16, and the present invention takes the value of 12.

[0082] See also Figure 2 , Figure 2 It is a schematic flow chart of a LSTM-based dialysis fistula maturity detection method provided in the second embodiment of the present application.

[0083] based on Figure 1 The embodiment shown, this embodiment Figure 2 As shown, the step S30 includes steps S21 to S22.

[0084] Step S21, the training module builds a preset number of LSTM units, each layer of LSTM units is followed by a Dropout layer, and the results are predicted through a Softmax layer to obtain a loss function;

[0085] Step S22: After obtaining the loss function, the loss function is optimized by the Adam optimization algorithm to find the global optimal point and introduce quadratic gradient correction.

[0086] Specifically, after the training module calculates the loss function through the predicted value, the loss function is reduced through the optimizer. The training module uses the Adam optimization algorithm to find the global optimal point and introduces quadratic gradient correction. The formula is as follows:

[0087] m t =μ*m t-1 +(1-μ)*g t

[0088]

[0089]

[0090]

[0091]

[0092] Where m t , n t They are the first-order moment estimate and the second-order moment estimate of the gradient, which can be regarded as the expected E|g t |, Moment estimates of ; It is for m t , n t , which can be approximated as an unbiased estimate of the expectation.

[0093] Based on the above embodiment, in this embodiment, step S40 includes:

[0094] The target sample is input into the sound classification model for testing, and the result is output as 0 and 1, where 0 represents normal fistula sound and 1 represents abnormal fistula sound.

[0095] See also Figure 3 , Figure 3 The embodiment of the present application provides a schematic block diagram of a LSTM-based dialysis fistula maturity detection system, which is used to execute the aforementioned LSTM-based dialysis fistula maturity detection method. The LSTM-based dialysis fistula maturity detection system can be configured on a server.

[0096] like Figure 3 As shown, the LSTM-based dialysis fistula maturity detection system 400 includes:

[0097] The acquisition module 10 is used to acquire sound samples and establish a sound sample database;

[0098] The feature extraction module 20 is used for preprocessing the sound sample and extracting features to obtain feature parameters;

[0099] A training module 30, used for training the training module based on a preset LSTM neural network and the characteristic parameters to establish a sound classification model;

[0100] The testing module 40 is used for testing the target sample through the sound classification model to provide a classification result.

[0101] Furthermore, the LSTM-based dialysis fistula maturity detection system includes:

[0102] A pre-emphasis module is used to perform pre-emphasis processing on the sound sample through a high-pass digital filter, and the preset transfer function of the high-pass digital filter is H(z)=1-az -1 , the relationship between the output of the pre-emphasis network and the input speech signal s(n) is:

[0103]

[0104] Where a is the pre-emphasis coefficient.

[0105] Furthermore, the LSTM-based dialysis fistula maturity detection system includes:

[0106] The short-time analysis module is used for analyzing the short-time stability of the sound sample by the feature extraction module using a movable finite-length window for weighted analysis, wherein the number of frames of the sound sample is f. n =(N-overlap) / inc=(N-wlen) / inc+1, wlen is the frame length, inc is the frame shift, N is the length of the voice data, and overlap is the overlapping part.

[0107] Furthermore, the LSTM-based dialysis fistula maturity detection system includes:

[0108] The window function module is used for the feature extraction module to perform windowing processing on the sound sample based on a preset window function, wherein the preset window function is:

[0109]

[0110] Furthermore, the feature extraction module 20 includes:

[0111] A Fourier transform unit is used to extract the characteristic parameters through a fast Fourier transform DFT, wherein the DFT is

[0112]

[0113] (n) is the input speech signal, and N represents the number of Fourier transform points;

[0114] The other extraction units are used to extract the characteristic parameters of the sound sample through a Mel filter bank, logarithmic operation, discrete cosine transform and / or dynamic feature extraction method.

[0115] Furthermore, the LSTM-based dialysis fistula maturity detection system includes:

[0116] A loss function generation module is used for the training module to build a preset number of LSTM units, each layer of LSTM units is followed by a Dropout layer, and the results are predicted through a Softmax layer to obtain a loss function;

[0117] The loss function optimization module is used to optimize the loss function through the Adam optimization algorithm after obtaining the loss function, find the global optimal point, and introduce quadratic gradient correction.

[0118] Furthermore, the testing module 40 includes:

[0119] The result output unit is used to input the target sample into the sound classification model for testing, and output the result as 0 and 1, 0 represents normal fistula sound, and 1 represents abnormal fistula sound.

[0120] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and each module described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0121] The above system can be implemented in the form of a computer program. Figure 4 Runs on the computer device shown.

[0122] See also Figure 4 , Figure 4 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device may be a server.

[0123] See also Figure 4 The computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0124] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any LSTM-based dialysis fistula maturity detection method.

[0125] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0126] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any LSTM-based dialysis fistula maturity detection method.

[0127] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 4The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0128] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0129] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0130] The acquisition module collects sound samples and establishes a sound sample database;

[0131] The feature extraction module preprocesses the sound sample and extracts features to obtain feature parameters;

[0132] The training module performs training based on the preset LSTM neural network and the characteristic parameters to establish a sound classification model;

[0133] The testing module tests the target sample through the sound classification model and provides a classification result.

[0134] In one embodiment, the feature extraction module pre-processes the sound sample to achieve:

[0135] The sound sample is pre-emphasized by a high-pass digital filter, and the preset transfer function of the high-pass digital filter is H(z)=1-az -1 , the relationship between the output of the pre-emphasis network and the input speech signal s(n) is:

[0136]

[0137] Where a is the pre-emphasis coefficient.

[0138] In one embodiment, the feature extraction module pre-processes the sound sample to achieve:

[0139] Based on the short-term stability of the sound sample, the feature extraction module performs a short-term analysis of the sound sample by using a weighted method using a movable window of finite length, wherein the number of frames of the sound sample is f n =(N-overlap) / inc=(N-wlen) / inc+1, wlen is the frame length, inc is the frame shift, N is the length of the voice data, and overlap is the overlapping part.

[0140] In one embodiment, the feature extraction module pre-processes the sound sample to achieve:

[0141] Based on a preset window function, the feature extraction module performs windowing processing on the sound sample, wherein the preset window function is:

[0142]

[0143] In one embodiment, feature extraction is performed to obtain feature parameters for achieving:

[0144] The characteristic parameters are extracted by fast Fourier transform DFT, wherein the DFT is

[0145]

[0146] (n) is the input speech signal, and N represents the number of Fourier transform points;

[0147] The characteristic parameters of the sound sample are extracted by Mel filter bank, logarithmic operation, discrete cosine transform and / or dynamic feature extraction method.

[0148] In one embodiment, the training module is trained based on a preset LSTM neural network and the feature parameters, before establishing a sound classification model, to achieve:

[0149] The training module builds a preset number of LSTM units, connects a Dropout layer after each LSTM unit, predicts the results through a Softmax layer and obtains a loss function;

[0150] After obtaining the loss function, the loss function is optimized by the Adam optimization algorithm to find the global optimal point and introduce quadratic gradient correction.

[0151] In one embodiment, the testing module tests the target sample through the sound classification model and provides a classification result for achieving:

[0152] The target sample is input into the sound classification model for testing, and the result is output as 0 and 1, where 0 represents normal fistula sound and 1 represents abnormal fistula sound.

[0153] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any LSTM-based dialysis fistula maturity detection method provided in the embodiments of the present application.

[0154] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc., equipped on the computer device.

[0155] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A LSTM-based method for detecting the maturity of dialysis fistulas. It is characterized in that The LSTM-based dialysis fistula maturity detection method is applied to the LSTM-based dialysis fistula maturity detection system. The LSTM-based dialysis fistula maturity detection system includes an acquisition module, a feature extraction module, a training module and a testing module. The LSTM-based dialysis fistula maturity detection method includes: The acquisition module collects sound samples and establishes a sound sample database; The feature extraction module preprocesses the sound sample and extracts features to obtain feature parameters; The training module performs training based on the preset LSTM neural network and the characteristic parameters to establish a sound classification model; The testing module tests the target sample through the sound classification model and provides a classification result; The feature extraction module pre-processes the sound sample, including: pre-emphasizing the sound sample by a high-pass digital filter, performing short-term analysis on the sound sample by weighting with a movable window of finite length, and performing windowing on the sound sample based on a preset window function; Among them, the training module is trained based on the preset LSTM neural network and the characteristic parameters. Before establishing the sound classification model, the training module builds a preset number of LSTM units, and each layer of LSTM units is followed by a Dropout layer. The results are predicted through the Softmax layer and the loss function is obtained.

2. The LSTM-based dialysis fistula maturity detection method according to claim 1, It is characterized in that The preset transfer function of the high-pass digital filter is H(z)=1-az -1 , the relationship between the output of the pre-emphasis network and the input speech signal s(n) is: Where a is the pre-emphasis coefficient.

3. The LSTM-based dialysis fistula maturity detection method according to claim 1, It is characterized in that Based on the short-term stability of the sound sample, the feature extraction module performs a short-term analysis of the sound sample by using a weighted method using a movable window of finite length, wherein the number of frames of the sound sample is f n =(N-overlap) / inc=(N-wlen) / inc+1, wlen is the frame length, inc is the frame shift, N is the length of the voice data, and overlap is the overlapping part.

4. The LSTM-based dialysis fistula maturity detection method according to claim 1, It is characterized in that The preset window function is:

5. The LSTM-based dialysis fistula maturity detection method according to claim 1, It is characterized in that The feature extraction to obtain feature parameters includes: The characteristic parameters are extracted by fast Fourier transform DFT, wherein the DFT is (n) is the input speech signal, and N represents the number of Fourier transform points; The characteristic parameters of the sound sample are extracted by Mel filter bank, logarithmic operation, discrete cosine transform and / or dynamic feature extraction method.

6. The LSTM-based dialysis fistula maturity detection method according to claim 1, It is characterized in that After obtaining the loss function, the loss function is optimized by the Adam optimization algorithm to find the global optimal point and introduce quadratic gradient correction.

7. The LSTM-based dialysis fistula maturity detection method according to any one of claims 1 to 6, It is characterized in that The testing module tests the target sample through the sound classification model and provides a classification result, including: The target sample is input into the sound classification model for testing, and the result is output as 0 and 1, where 0 represents normal fistula sound and 1 represents abnormal fistula sound.

8. A detection system for the LSTM-based dialysis fistula maturity detection method according to any one of claims 1 to 7, It is characterized in that include: A collection module, used for collecting sound samples and establishing a sound sample database; A feature extraction module is used for preprocessing the sound sample and extracting features to obtain feature parameters; A training module, used for training the training module based on a preset LSTM neural network and the characteristic parameters to establish a sound classification model; A testing module, used for the testing module to test the target sample through the sound classification model and provide a classification result; The feature extraction module pre-processes the sound sample, including: pre-emphasizing the sound sample by a high-pass digital filter, performing short-term analysis on the sound sample by weighting with a movable window of finite length, and performing windowing on the sound sample based on a preset window function; Among them, the training module is trained based on the preset LSTM neural network and the characteristic parameters. Before establishing the sound classification model, the training module builds a preset number of LSTM units, and each layer of LSTM units is followed by a Dropout layer. The results are predicted through the Softmax layer and the loss function is obtained.

9. A computer device, It is characterized in that The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the LSTM-based dialysis fistula maturity detection method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the LSTM-based dialysis fistula maturity detection method according to any one of claims 1 to 7.